Simulation Approaches in Plant Breeding. Jiankang Wang

Transcription

1 Simulation Approaches in Plant Breeding Jiankang Wang 1

2 Outlines for presentation Conventional quantitative genetics Plant breeding and quantitative genetics QTL mapping Background of breeding simulation QuLine tool for genetics and breeding QuLine applications Demonstration and Hands-on 2

3 Background on breeding simulation 3

4 Toluca (1000 crosses made from 200 parents) A x B About 30% of the crosses are discarded in F1 PLANT BREEDING IS A COMPLEX PROCESS WITH MANY DECISIONS About 40% of crosses retained after F7 About 20% of crosses retained after two rounds of yield trials Cd. Obregon (Each selected F1 harvested in bulk) F1 Toluca (30-80 plants harvested individually for MODPED) F2 (30-80 plants harvested in bulk for SELBLK) Cd. Obregon (Bulk of 10 spikes for MODPED) F3 (Bulk of 30 spikes for SELBLK) Toluca (Bulk of 10 spikes for MODPED) F4 (Bulk of 30 spikes for SELBLK) Cd. Obregon (Bulk of 10 spikes for MODPED) F5 (Bulk of 30 spikes for SELBLK) Toluca (10 plants harvested individually for MODPED) F6 (40 plants harvested individually for SELBLK) Cd. Obregon (Bulk of whole plot) F7 Toluca and El Batan (Bulk of whole plot) F8 field tests Cd. Obregon (Bulk of whole plot) F8 yield trial F8 small plot evaluation Toluca and El Batan (Bulk of whole plot) F9 field tests About 1 to 3% of crosses will derive lines that can be used as cultivars (Reserve seed ) (Reserve seed ) Cd. Obregon (Bulk of whole plot) F9 yield trial F9 small plot evaluation Toluca and El Batan (Bulk of whole plot) F10 stripe rust screening F10 leaf rust screening International screening nursery and yield trial 4

5 Why do we need tools in breeding? To improve the efficiency of traditional phenotypic selection through exploring various options To avoid the simplified assumptions made in classical quantitative genetic theory To better use the large amount of gene information available from Genomics research January 1992: 59,317 datasets, 77,805,556 bp March 2005: 43,118,204 datasets, 47,009,081,750 bp QTL mapping (CAB, April 2005) 3497 publications on QTL mapping 1581 publications in plants 5

6 Why do we need tools in breeding? To build a bridge between the biological data and breeders requirements To combine all these sources of data into knowledge that breeders can use in their breeding programs 6

7 Questions that can be studied by QuLine: A genetic and breeding simulation tool 1. Comparison of breeding efficiencies from different selection strategies and their modifications. Which breeding method should be adopted? 2. Balance between the number of crosses and population size of segregating generations. What shall the breeder do if the available resources increase or reduce? 3. Evaluation of marker-assisted selection (MAS). When and how should MAS be used? 4. Comparative value of single, top, back, and double crosses in breeding?

8 Questions that can be studied by QuLine (mainly for inbred line development) 5. The correlation between parents and their offspring. Can F1 or F2 hybrids predict the performance of their advanced lines? For early generation selection, how early is too early? 6. When to use DH (doubled haploid)? Early generation: many individuals need to be tested Advanced line stage: lines will be good for other traits, but the desired genotype may be lost during selection due to population size, trait associations and genetic drift

9 QuLine tool for genetics and breeding 9

10 Available breeding simulation tools QuLine, a computer software that simulates breeding programs for developing inbred lines QuHybrid, a computer software that simulates breeding programs for developing hybrids QuMARS, a computer software that simulates marker-assisted recurrent selection and genome-wide selection

11 Test cross implemented in QuHybrid Family PopBefore(1) PopBefore(2) PopBefore(3) Testers Family size F11 F12 F13 F14 F15 F21 F22 F23 F31 F32 F33 F34 F35 F36 T1 T2 T3 Making testcrosses Testcross 1 F11 T1 F12 T1 F13 T1 F14 T1 F15 T1 F21 T1 F22 T1 F23 T1 F31 T1 F32 T1 F33 T1 F34 T1 F35 T1 F36 T1 Testcross 2 F11 T2 F12 T2 F13 T2 F14 T2 F15 T2 F21 T2 F22 T2 F23 T2 F31 T2 F32 T2 F33 T2 F34 T2 F35 T2 F36 T2 Testcross 3 F11 T3 F12 T3 F13 T3 F14 T3 F15 T3 F21 T3 F22 T3 F23 T3 F31 T3 F32 T3 F33 T3 F34 T3 F35 T3 F36 T3 Among family selection Testcross 1 F11 T1 F12 T1 F13 T1 F14 T1 F15 T1 F21 T1 F22 T1 F23 T1 F31 T1 F32 T1 F33 T1 F34 T1 F35 T1 F36 T1 Testcross 2 F11 T2 F12 T2 F13 T2 F14 T2 F15 T2 F21 T2 F22 T2 F23 T2 F31 T2 F32 T2 F33 T2 F34 T2 F35 T2 F36 T2 Testcross 3 F11 T3 F12 T3 F13 T3 F14 T3 F15 T3 F21 T3 F22 T3 F23 T3 F31 T3 F32 T3 F33 T3 F34 T3 F35 T3 F36 T3 Within family selection F21 T1 F22 T1 F23 T1 F21 T2 F22 T2 F23 T2 F21 T3 F22 T3 F23 T3 F31 T1 F32 T1 F33 T1 F34 T1 F35 T1 F36 T1 F31 T2 F32 T2 F33 T2 F34 T2 F35 T2 F36 T2 F31 T3 F32 T3 F33 T3 F34 T3 F35 T3 F36 T3 Population after selection, this population will be used as the starting for the next generation Family PopAfter(1) PopAfter(2) PopAfter(3) Family size F22 F23 F31 F33 F34 F36

12 Hybrids Stage I: Testcrossing: Starting from a single cross, To generate the initial population for recurrent selection; To build the prediction equation of breeding traits on markers P1 P2 The flowchart of QuMARS F1 200 F2 individuals 200 F2:3 families Genotyping each family using bulk DNA of 10 individuals individuals each family Testcrossing between each F2:3 family and the tester 30 selected F2:3 families based on testcross performance 200 testcross families Phenotyping each testcross family Selecting using MARS or GWS Stage II: Recurrent selection: To intermate the selected families (S0); To grow and genotype randomly mated progenies; To select individuals based on MARS or GWS Cycle 0 Intermating of the 30 selected F2:3 families Cycle individuals Genotyping each individual Selecting using MARS or GWS 30 selected individuals Intermating of the 30 selected findividuals Cycle n 300 individuals Genotyping each individual Selecting using MARS or GWS 30 selected individuals End of the recurrent selection Stage III: Inbred development: To generate inbred lines through pedigree breeding Repeated self-pollination to derive inbred lines Inbred lines Note: QuMARS stops here Repeated testcrossing evaluation to derive hybrids

13 What can QuLine do? Comparison of genetic gains from different selection methods Change in population mean Change in gene frequency Change in Hamming distance (distance of a selected genotype to the target genotype) Comparison of cross performance Selection history Rogers genetic distance Number of lines retained from each cross Comparison of cost efficiency Number of families Individual plants per generation Validation of theories 13

14 In genetics (implemented by the QU-GENE engine) Most genetic phenomena, if not all, can be defined in the QU-GENE engine input file (QUG). Among them are: Multiple alleles (e.g. Glutenin genes in wheat) Linkage (between gene and marker, between genes, between markers) Additive, dominance and epistasis Pleiotropy (one gene effects multiple traits) Genotype by environment interaction Molecular markers (dominant, or co-dominant) 14

15 In breeding (implemented by the QuLine module) Most, if not all, breeding methods for selfpollinated crops, can be defined and then simulated in QuLine. Among them are: Pedigree system (including SSD) Bulk-population Doubled haploid Marker-assisted selection (include marker-based selection) Recurrent selection within one population Many modifications and combinations 15

16 How does QuLine work? Two input files are needed QUG file containing the necessary information for a genotype and environment (GE) system and initial population(s) of genotypes. It is the input for the QU- GENE engine. Two kinds of output files will be generated from the engine. GES file for defining a GE system (= input for QuLine) POP file for defining the initial population (= input for QuLine) QMP file containing the necessary information for the breeding strategies to be simulated (e.g. pedigree, bulk, SSD, DH, etc.) (= input for QuLine) 16

17 General procedure for crossing and selection in F1 POP file from the engine or the previous breeding cycle Propagation of CB Making crosses Among-family selection Within-family selection Generation advance Popbulk(:) for bulk 17

18 18 Popbulk(:) for bulk Propagation process Among-family selection Generation advance Within-family selection Popbulk(:) for pedigree General procedure for propagation and selection in F2 and onwards

19 Define the QMP file for the selected bulk selection method: an example Define the breeding method in a way that the computer can understand 19

20 General simulation parameters Number of runs: any integer Number of breeding cycles: any integer Number of crosses in F1 generation: any integer Indicator for crossing block update, 0 or 1. 0: All individuals after a breeding cycle are used as the parents for the next cycle; 1: The best individuals in the final selected population and the initial crossing block are selected as the parents for next cycle 20

21 General simulation parameters Indicator for outputting GE system details, 0 for no output, and 1 for output Indicator for outputting population details, 0 for no output, and 1 for output Indicator for outputting selection history, 0 for no output, and 1 for output Indicator for outputting Rogers distance for each cross and the lines retained from each cross, 0 for no output, and 1 for output Indicator for outputting correlation coefficient, 0 for no output, and 1 for output 21

22 The number of models in the GE system and the number of runs for breeding strategy Four nested loops in QuLine Loop for all models All random effects in the GE system will be assigned values Determine the parents for all crosses for the first cycle Loop for all runs Loop for all strategies»loop for all cycles All strategies start from the same point (same crossing block and same crosses), so that they can be properly compared. 22

23 Parameters to describe a set of breeding strategies Number of strategies For each strategy Strategy name: any character Number of generations in the strategy: any integer more than 0 Definition for each generation 23

24 Definition of a generation Number of selection rounds in the generation: any integer more than 1 Seed source indicator 0: Seed for selection round i (i > 1) come from round 1 1: Seed for selection round i (i > 1) come from round i-1 Definition of each selection round 24

25 An example for seed source indicator 0 Practical breeding F6 field test at Toluca Virtual breeding F6 Families selected ,760 F7 field test at Cd. Obregon F8 field test at Toluca and El Batan F8 yield trial and F8 small plot evaluation at Cd. Obregon F7 F8 (T), F8 (B) F8 (YT) F8 small plot Round 1 for F7 14,760 3,868 Rounds 2 and 3 3,868 2,163 1,974 Round 4 1,

26 An example (LRC, Toowoomba, Australia) for seed source indicator 1 Pedigree in F2 Families selected Bulk in F3 Round 1 for F PYEI, F4 Round PYEII, F5 Round Pedigree in F

27 Definition of each selection round Title for the generation Seed propagation type (within a family) clone, asexual reproduction DH, doubled haploid self, self-pollination backcross, backcrossed to one parent topcross, crossed with a third parent (three-way cross) doublecross, crossed with another F1 random, random mating noself, random mating but self-pollination is eliminated 27

28 Definition of each selection round Generation advance method (or harvest method): management of the selected individual plants in a family pedigree: the selected plants in each family are harvested individually, resulting a few families in the next generation bulk: the selected plants in each family are harvested in bulk, resulting one family in the next generation 28

29 Definition of each selection round Field experiment design Number of replications for each family Number of plants in each replication Number of test locations Environment type for each test location defined in the GE system if 0, randomly determined based on environment frequency 29

30 Definition of each selection round Among family and within family selection Number of traits used for selection Definition of each trait 30

31 Definition of each trait used in selection Trait number, for the trait in selection (0 when marker score is used in selection) Selection mode T for top, e.g. yield, tillering, grains per spike and 1000-kernel weight B for bottom, e.g. lodging and rusts M for middle, e.g. height and heading R for random, for some special studies TV for top value BV for bottom value TN for a number of individuals/families with top phenotypic values BN for a number of individuals/families with bottom phenotypic values RN for a number of individuals/families to be selected randomly Selected proportion or value: the proportion or value of individual plants in a family (for within family selection) or of families (for among family selection) to be selected Proportion selection Threshold selection Number selection 31

32 An example of generation definition Rounds of selection Seed source indicator Generation title Seed propagation type Generation advance method Replicat ions Plot size Test locations Environment type 1 0 F6 self pedigree , Toluca 4 0 F7 self bulk , Obregon F8(T) self bulk , Toluca F8(B) self bulk , El Batan F8(YT) self bulk , Obregon 1 0 F8(SP) self bulk , Obregon 32

33 Trait Selection mode Traits, their selection modes and Yield selected proportions Lodging Stem rust Leaf rust Stripe rust Height Tillering Heading Grains per spike T B B B B M T M T T 1000 kernel weight F6, among F6, within F7, among F8(T), among F8(B), among F8(YT), among F8(SP), among Total

34 In QuLine, a breeding program looks like!nr SS GT PT GA RP PS NL ET... Row 1! AT (ID SP SM)... Row 2! WT (ID SP SM)... Row CB self bulk F1 singlecross bulk B B B M T B T F2 self pedigree B B B M T B T B B B M T B T T F3 self bulk B B B M T B T B M B T T F4 self bulk B B M T B T B M B T T F5 self bulk B B M T B T B M B T T F6 self pedigree B B M T B T B B T B T F7 self bulk B B M T B T T AL(T) self bulk B B M T B T AL(B) self bulk B PYT self bulk T

35 Steps to run QuLine Input information about the GxE system and populations QUGENE Breeding strategies GE system Population QuLine Major outputs from QuLine.cro.fit.fix.fre.his.ham.pou.rog Crosses after selection Genetic advance Genes fixed Gene frequency Selection history Hamming distance Population details Cross details.var Variance components 35

36 Comparison of two breeding strategies: modified pedigree (MODPED) and selected bulk (SELBLK) Crop Science, 2003, 43: Crop Science, 2004, 44:

37 Dr. Borlaug and the Green Revolution Mexican-Rockefeller Foundation Agricultural Program, 1944 Mexico became self-sufficient in most food commodities by the late 1950s Young scientists were trained in Mexico Semidwarf wheats (Rht1 gene and Norin 10) , wheat and rice production had increased by 50% Awarded Nobel Peace Prize in

38 CIMMYT s Shuttle Breeding November to May Cd. Obregon, 27ºN, 39 masl t/ha under irrigation; 1-2t/ha under reduced irrigation Mexico City May to November Toluca, 19ºN, 2640 masl. High rainfall ( mm) El Batan 38

39 Breeding methods with self-pollinated crops Allard, R.W Principles of plant breeding. John Wiley & Sons, Inc. Stoskopf, N.C Plant Breeding Theory and Practice. Westview Press. Mass and pure-line selection The pedigree system The bulk population method The backcross breeding method Single seed descent (SSD), a special case of pedigree system Recurrent selection breeding method Mutation breeding Haploid breeding system (doubled haploid) 39

40 Breeding methods in CIMMYT s wheat breeding program Pedigree system: before Pedigree selection is used from F2 to F6. Modified pedigree/bulk (MODPED): in /94. Pedigree selection is used in F2 and F6, and bulk selection is used in other generations. Selected bulk (SELBLK): after Pedigree selection is used only in F6, and bulk selection is used in other generations. 40

41 Toluca (1000 crosses made from 200 parents) A x B Germplasm flow for simple crosses made in Toluca and targeted to ME1 MODPED: modified pedigree/bulk SELBLK: selected bulk methods Cd. Obregon (Each selected F1 harvested in bulk) F1 Toluca (30-80 plants harvested individually for MODPED) F2 (30-80 plants harvested in bulk for SELBLK) Cd. Obregon (Bulk of 10 spikes for MODPED) F3 (Bulk of 30 spikes for SELBLK) Toluca (Bulk of 10 spikes for MODPED) F4 (Bulk of 30 spikes for SELBLK) Cd. Obregon (Bulk of 10 spikes for MODPED) F5 (Bulk of 30 spikes for SELBLK) Toluca (10 plants harvested individually for MODPED) F6 (40 plants harvested individually for SELBLK) Cd. Obregon (Bulk of whole plot) F7 Toluca and El Batan (Bulk of whole plot) F8 field tests (Reserve seed ) Cd. Obregon (Bulk of whole plot) F8 yield trial F8 small plot evaluation Toluca and El Batan (Bulk of whole plot) F9 field tests (Reserve seed ) Cd. Obregon (Bulk of whole plot) F9 yield trial F9 small plot evaluation Toluca and El Batan (Bulk of whole plot) F10 stripe rust screening F10 leaf rust screening International screening nursery and yield trial 41

42 Trait, segregating gene number, gene effects and trait heritability Trait Genes Gene effect type AA Aa aa Trait range h b 2 (Indiv. plant) Yield 20, 40 E0, E1, E2 Random value from UD (0, 1) 0.05 Lodging 3 additive Stem rust 5 additive Leaf rust 5 additive Yellow rust 5 additive Height 3 additive Tillers/plant 3 additive Heading 5 additive Grains/spike 5 additive Seed weight 5 additive

43 Trait correlation and pleiotropy Trait Yield Lodging Stem rust Leaf rust Yellow rust Height Tillers/ plant Heading Grains/ spike Seed weight Yield Lodging Stem rust Leaf rust Estimated by CIMMYT breeders Yellow rust Height Tillers/plant Heading 0.60 Estimated from the defined genetic model Grains/spike Seed weight

44 Experiment design 12 Genotype and environment (GE) systems Two yield gene numbers: 20 and 40, two alleles for each gene Pleiotropy (same gene effects various traits): absent and present Epistasis (multiple gene interaction): no epistasis, digenicepistasis, and tri-genic epistasis Linkage: no linkage (independent gene segregation) Initial population 200 parents, gene (allele) frequencies of 0.5 for all genes 1000 crosses were made 258 lines were selected after 10 generations of selection 44

45 % of target genotype Result 1: Genetic gain in yield from SELBLK is 3.3% higher than MODPED. SELBLK is slightly more efficient SELBLK MODPED 50 Initial Population Final selected 45

46 Genetic gain (%) For gains per spike and 1000-kernel weight, SELBLK has a faster genetic gain. For tillers/plant, MODPED has a faster genetic gain SELBLK MODPED Yield Tillers/plant GRN/SPK TKW Lodging tol. SR res. Traits LR res. YR res. Height Heading 46

47 F1 F2 F3 F4 F5 F6 F7 F8(T) F8(B) F8(YT) F8(SP) F9(T) F9(B) F9(YT) F9(SP) F10(YR) F10(LR) Number of crosses Result 2: SELBLK retained 25% more crosses in the final selected population (more genetic diversity retained) SELBLK MODPED Generation 47

48 F1 F2 F3 F4 F5 F6 F7 F8(T) F8(B) F8(YT) F8(SP) F9(T) F9(B) F9(YT) F9(SP) F10(Y R) F10(LR) Number of individual plants Result 3: SELBLK required 1/3 less land from F1 to F8 than MODPED. SELBLK is more cost-effective SELBLK MODPED Generations 48

49 F1 F2 F3 F4 F5 F6 F7 F8(T) F8(B) F8(YT) F8(SP) F9(T) F9(B) F9(YT) F9(SP) F10(Y R) F10(LR) Number of families Result 4: SELBLK produced 40% less families (plots) to be planted from F1 to F8 (less labor required) SELBLK MODPED Generation 49

50 Modeling of the Single Backcrossing Breeding Strategy (SBBS) Theor. Appl. Genet., 2009, 118:

51 Ravi Singh, wheat breeder and pathologist in CIMMYT Adapted wheats X durable disease resistance donors 1-2 times of backcrossing with the adapted Large BCF2 population Select adapted lines combined with durable resistances

52 Estimated percentages of favourable alleles or gene combinations in different parental lines in Category favorable genes (%) Example % total lines Elite adapted lines (EAL) Major released cultivars in targeted megaenvironments 10 (MEs) either developed by CIMMYT or by partners Adapted lines (AL) Elite advanced lines from CIMMYT s 60 International Nursery and Yield Trials Intermediate adapted lines Advanced lines from CIMMYT s Yield Trials in 10 (IAL) Ciudad Obregón and Toluca, Mexico Un-adapted (or nonadapted) Land races 2 lines (UAL) Second generation of resynthesized Derived lines between the first generation of 10 wheat (SYNII) re-synthesized wheat derivatives and adapted lines First generation of re Derived lines between primary re-synthesized 5 synthesized wheat (SYNI) Primary re-synthesized wheat (SYN0) wheat breeding at CIMMYT wheat and adapted lines 0-30 Inter-specific crosses between Triticum durum and Aegilops tauschii 3

53 Two traits defined in QU-GENE Adaptation 200 genes on the 21 wheat chromosomes Lowest adaptation with no favorable alleles: 0 Highest adaptation with all favorable alleles: 100 Heritability: 0.5 Donor traits (DT) to be transferred 10 genes governing the donor traits Lowest DT with no favorable alleles: 0 Highest DT with all favorable alleles: 10 Heritability: 0.5

54 Adapted parental groups in simulation Gene frequency of favorable adaptation alleles fixed at 0.8 A0: the frequency of favorable DT alleles is 0 A2: the frequency of favorable DT alleles is 0.2 A4: the frequency of favorable DT alleles is 0.4 A6: the frequency of favorable DT alleles is 0.6 A8: the frequency of favorable DT alleles is 0.8

55 Donor parental groups in simulation Gene frequency of favorable DT alleles fixed at 1.0 D0: the frequency of favorable adaptation alleles is 0 D1: the frequency of favorable adaptation alleles is 0.1 D2: the frequency of favorable adaptation alleles is 0.2 D3: the frequency of favorable adaptation alleles is 0.3 D4: the frequency of favorable adaptation alleles is 0.4 D5: the frequency of favorable adaptation alleles is 0.5 D6: the frequency of favorable adaptation alleles is 0.6 D7: the frequency of favorable adaptation alleles is 0.7

56 Category Crosses made between different parental groups for wheat breeding at CIMMYT Percentage of total crosses Similarity to defined parental groups (EAL+AL) (EAL+AL) 65 (A0+A2+A4+A6+A8) D7 (EAL+AL) IAL 10 (A0+A2+A4+A6+A8) D5+D6) (EAL+AL) UAL 5 (A0+A2+A4+A6+A8) (D2+D3+D4) (EAL+AL) SYNII 10 (A0+A2+A4+A6+A8) (D6+D7) (EAL+AL) SYNI 7 (A0+A2+A4+A6+A8) (D4+D5) (EAL+AL) SYN0 3 (A0+A2+A4+A6+A8) (D0+D1+D2)

57 The single backcrossing breeding strategy (SBBS) Generation Seed propagation method No. crosses or families grown F 1 B 1 F 1 Hand pollination between adapted and donor lines Backcrossing to the adapted parents Individuals per cross or family No. selected crosses or families B 1 F 2 Selfing B 1 F 3 Selfing B 1 F 4 Selfing B 1 F 5 Selfing B 1 F 6 Selfing Final selected advanced lined 10 No. selected individuals in each cross or family

58 Four crossing strategies defined in QuLine No backcross One time of backcross Two times of backcross Three times of backcross Generation advance method B0 B1 B2 B3 F 1 F 1 F 1 F 1 Bulk F 2 BC 1 F 1 BC 1 F 1 BC 1 F 1 Bulk F 3 BC 1 F 2 BC 2 F 1 BC 2 F 1 Bulk F 4 BC 1 F 3 BC 2 F 2 BC 3 F 1 Bulk F 5 BC 1 F 4 BC 2 F 3 BC 3 F 2 Bulk F 6 BC 1 F 5 BC 2 F 4 BC 3 F 3 Pedigree F 7 BC 1 F 6 BC 2 F 5 BC 3 F 4 Bulk

59 Six selection schemes AD: Adaptation is selected first, followed by the selection for DT DA: DT is selected first, followed by the selection for Adaptation ADA: Adaptation is selected first, followed by the selection for DT, and adaptation is selected again DAD: DT is selected first, followed by the selection for Adaptation, and DT is selected again ADAD: Adaptation and DT are selected two times in each generation, and adaptation is selected first DADA: Adaptation and DT are selected two times in each generation, and DT is selected first

60 % highest donor trait % highest adaptation Genetic advance of selection scheme AD A B B0 B1 B2 B3 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 D0D1D2D3D4D5D6D7 A0 A2 A4 A6 A8 Donor and adapted parental groups

61 Conclusions We recommend the use of SBBS based on three assumptions: multiple genes governing the phenotypic traits to be transferred from donor parents to adapted parents donor parents still have some favorable genes that may contribute to the improvement of adaptation in the recipient parents even under low adaptation the conventional phenotypic selection is applied or the individual genotypes cannot be precisely indentified

62 Breeding with known gene information Aust. J. Agric. Res., 2005, 56:

63 Glutenin genes and wheat quality Six glutenin genes Glu-A1 (1AL), Glu-B1 (1BL), Glu-D1 (1DL) for HMW Glu-A3 (1AS), Glu-B3 (1BS), Glu-D3 (1DS) for LMW Two end-use quality traits commonly used by wheat breeders Rmax (BU), for maximum dough resistance Extensibility (cm), for dough extensibility Multiple alleles on each gene locus Glu-A1: 1, 2, and Null Glu-B1: 7, 7+8, 7+9, 6+8, 20, 13+16, 14+15, 17+18, and Glu-D1: 2+12, 4+12, 5+10, and 2+T2 63

64 Selected parents Parent Rmax Extensibility Silverstar Silverstar Silverstar Silverstar Westonia Krichauff Machete Diamondbird

65 Four selection schemes Parent 1 Parent 2 F1 F2: 1000 individuals F8: 1000 lines through SSD R0.04 R0.2E0.2 E0.2R0.2 E0.04 Trait to be Lines Trait to be Lines Trait to be Lines Trait to be Lines selected selected selected selected selected selected selected selected Step 1 Rmax 40 Rmax 200 Extensibility 200 Extensibity 40 Step 2 n.a. Extensibility 40 Rmax 40 n.a. 65

66 Rmax (BU) Extensibility (cm) (a) Westonia (b) (c) Krichauff (d) (e) Machete (f) Diamondbird (g) (h) R0.04 R0.2E0.2 E0.2R0.2 E0.04 R0.04 R0.2E0.2 E0.2R0.2 E0.04 Selection scheme (trait followed by selected proportion) 66

67 The best sister lines under each breeding objective and selection scheme Parent Objective R0.04 R0.2E0.2 E0.2R0.2 E0.04 Westonia High Rmax 3, 7 3, 7 3, 7 1, 3 High Ext. 1 1, 5 1, 3, 5 1,3,5,7 Krichauff High Rmax 1,3,5,7 1,3,5,7 3, 7 3, 7 High Ext. 1,3,5,7 1,3,5,7 1, 5 1, 5 Machete High Rmax 3,4,7,8 3,4,7,8 4, 8 None High Ext. 1,2,5,6 1,2,5,6 1, 2, 3 1,2,3,4 Diamondbird High Rmax 1,2,3,4 1, 3, 4 3, 4 3, 4 High Ext. None None 1,2,5,6 1,2,5,6 67

68 Efficient selection of multiple genes via marker-assisted selection, an example in wheat Wang, J., S.C. Chapman, D.B. Bonnett, G.J. Rebetzke, and J. Crouch Application of population genetic theory and simulation models to efficiently pyramid multiple genes via marker-assisted selection. Crop Science 47:

69 Nine major genes to be pyramided in wheat Gene Rht-B1 Rht-D1 Rht8 Sr2 Cre1 VPM Glu-B1 Glu-A3 tin Chr. 4BS 4DS 2DL 3BS 2BL 7DL 1BL 1AS 1AS Marker Codom Codom Codom Codom Dom Dom Codom Codom Codom MK-gene distance HM14BS Rht-B1a Rht-D1a Rht8 sr2 cre1 vpm Glu-B1a Glu-A3e Tin Sunstate Rht-B1a Rht-D1b rht8 Sr2 cre1 VPM Glu-B1i Glu-A3b Tin Silverstar+ tin Rht-B1b Rht-D1a rht8 sr2 Cre1 vpm Glu-B1i Glu-A3c tin Target Rht-B1a Rht-D1a Rht8 Sr2 Cre1 VPM Glu-B1i Glu-A3b tin

70 One strategy identified by QuLine to combine the nine genes from topcross Selection of Sunstate as the final parent (having largest number of favorable alleles) in the topcross Stage I: Selection for Rht-B1a and Glu-B1i homozygotes, and enrichment of rht8, Cre1, and tin in TCF1 Stage II: Selection of homozygotes for one target allele, e.g. Rht8, and enrich remaining target alleles in TCF2 Stage III: Selection of the target genotype in DHs/RILs

71 Comparison with other strategies For this strategy, one target genotype can be selected by screening < 600 individuals/lines For one-stage selection in advanced generations, one target genotype can be selected be screening > 3500 lines For one-stage selection in early generations, say TCF2, one target genotype can be selected be screening millions of individuals

72 Using QTL mapping results to design the breeding program Genetical Research, 2006, 88: Theor. Appl. Genet., 2007, 115:

73 M3 M6 M10 M12 M14 M17 M23 M25 M26 M30 M34 M60 M73 M1 M10 M12 M14 M19 M23 M35 M46 QTL for grain length and grain width in rice using 65 CSS lines Grain length Grain width LOD 值 4 LOD 值

74 基因型预测值 基因型预测值 Prediction using identified QTL 0.5 粒长 粒宽 0.3 r = r = 粒长观测值 (mm) 粒宽观测值 (mm) 74

75 基因型效应的预测值 Designing the target genotype (assuming large seed is favored) Target genotype 2 粒长 粒宽 最短粒长型最长粒长型最短粒宽型最长粒宽型设计基因型 75

76 Choosing parental lines Genotype M1 M6 M12 M23 M25 Prediction in GL (mm) GW (mm) Longest grain Widest grain Designed TG CSSL CSSL CSSL

77 Achieving the designed TG Three option for crossing TC1:(CSSL5 CSSL16) CSSL19 TC2:(CSSL5 CSSL19) CSSL16 TC3:CSSL5 (CSSL16 CSSL19) Two MAS schemes (just for example) Scheme1:MAS in F8 only Scheme2:MAS in F2 and F8 77

78 Choosing the best crossing and selection schemes TC MAS Selected Ind. In F2 F8 families before MAS F8 families after MAS (S.E.) DNA samples TC1 Scheme (3.27) Scheme (3.37) TC2 Scheme (7.06) Scheme (7.16) TC3 Scheme (5.45) Scheme (5.14) DNA samples per TG 78

79 Frequency (%) Frequency (%) Frequency (%) Frequency distribution of the number of DG1 from 1000 simulations (Breeding is science, art and chance!) 50 A. TC1: CSSL4/CSSL28//CSSL Scheme 1 Scheme 2 Scheme B. TC2: CSSL4/CSSL49//CSSL C. TC3: CSSL28/CSSL49//CSSL Number of target genotypes selected

80 Breeding is a tedious and long procedure when considering the nature of complex inheritance: linkage, epistasis, pleiotropy, GbyE, etc. Dr. Borlaug spent more than 20 years from this tall wheat to this semi-dwarf wheat, which triggered the Green Revolution! 80

81 Acknowledgements Grain Research and Development Corporation (GRDC), Australia ( ) Generation Challenge Program (GCP) of CGIAR ( ) HarvestPlus Challenge Program of CGIAR ( ) National 973 Programs of China for Key Basic Scientific Research ( ) National 863 Programs of China for High Technological Research ( ) Natural Science Foundation of China ( )

82 Demonstration of QuLine 82